Research map / Probabilistic and scientific modeling / Molecules and proteins
Can Information Flows Suggest Targets for Interventions in Neural Circuits?
Praveen Venkatesh, Sanghamitra Dutta, Neil Mehta, Pulkit Grover
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| prune edges by bias-to-accuracy information flow ratio | small ANN · Synthetic | bias and accuracy: 10 percentage points bias vs 2 percentage points accuracy | Improved |
| prune edges by accuracy-to-bias flow ratio (ablation) | small ANN · Synthetic | bias | Hurt |
| prune nodes instead of individual edges (6 edges each) (ablation) | small ANN · Modified Adult | bias-accuracy tradeoff | Improved |
| prune edges in second layer instead of first layer (ablation) | small ANN · Synthetic | accuracy/bias change | Hurt |
| prune by accuracy-to-bias flow ratio (ablation) | small ANN · Modified Adult | accuracy and bias | Mixed |
| path-based pruning by accuracy-to-bias flow ratio (ablation) | ANN · Modified Adult | bias | Improved |
| random pruning vs flow-ratio-based pruning (ablation) | small ANN · Modified Adult | bias-accuracy tradeoff | Improved |
| scale to deep MNIST ANN, per-edge pruning | deep ANN with 5 hidden layers · MNIST | dependence of effect on information flow: 0.12 | Hurt |
| increase network size | information flow pruning · MNIST | number of units to prune | Hurt |
| final-layer edges versus first-layer edges (ablation) | information flow dependence · Modified Adult large ANN | Pearson correlation: 0.57 vs 0.04 | Hurt |
Closest papers by meaning
- Preferential Normalizing Flows (NeurIPS 2024 · similarity 0.84)
- Causal Fairness under Unobserved Confounding: A Neural Sensitivity Framework (ICLR 2024 · similarity 0.83)
- Rethinking Fair Representation Learning for Performance-Sensitive Tasks (ICLR 2025 · similarity 0.82)
- Everything, Everywhere, All at Once: Is Mechanistic Interpretability Identifiable? (ICLR 2025 · similarity 0.82)
- Pathways of Visual Information Flow in Vision-Language Models (NeurIPS 2026 · similarity 0.82)
- InfoNet: Neural Estimation of Mutual Information without Test-Time Optimization (ICML 2024 · similarity 0.82)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
